[GitHub/mt8127/android_kernel_alcatel_ttab.git] / net / core / datagram.c

/*
 *	SUCS NET3:
 *
 *	Generic datagram handling routines. These are generic for all
 *	protocols. Possibly a generic IP version on top of these would
 *	make sense. Not tonight however 8-).
 *	This is used because UDP, RAW, PACKET, DDP, IPX, AX.25 and
 *	NetROM layer all have identical poll code and mostly
 *	identical recvmsg() code. So we share it here. The poll was
 *	shared before but buried in udp.c so I moved it.
 *
 *	Authors:	Alan Cox <alan@lxorguk.ukuu.org.uk>. (datagram_poll() from old
 *						     udp.c code)
 *
 *	Fixes:
 *		Alan Cox	:	NULL return from skb_peek_copy()
 *					understood
 *		Alan Cox	:	Rewrote skb_read_datagram to avoid the
 *					skb_peek_copy stuff.
 *		Alan Cox	:	Added support for SOCK_SEQPACKET.
 *					IPX can no longer use the SO_TYPE hack
 *					but AX.25 now works right, and SPX is
 *					feasible.
 *		Alan Cox	:	Fixed write poll of non IP protocol
 *					crash.
 *		Florian  La Roche:	Changed for my new skbuff handling.
 *		Darryl Miles	:	Fixed non-blocking SOCK_SEQPACKET.
 *		Linus Torvalds	:	BSD semantic fixes.
 *		Alan Cox	:	Datagram iovec handling
 *		Darryl Miles	:	Fixed non-blocking SOCK_STREAM.
 *		Alan Cox	:	POSIXisms
 *		Pete Wyckoff    :       Unconnected accept() fix.
 *
 */

#include <linux/module.h>
#include <linux/types.h>
#include <linux/kernel.h>
#include <asm/uaccess.h>
#include <asm/system.h>
#include <linux/mm.h>
#include <linux/interrupt.h>
#include <linux/errno.h>
#include <linux/sched.h>
#include <linux/inet.h>
#include <linux/netdevice.h>
#include <linux/rtnetlink.h>
#include <linux/poll.h>
#include <linux/highmem.h>
#include <linux/spinlock.h>

#include <net/protocol.h>
#include <linux/skbuff.h>

#include <net/checksum.h>
#include <net/sock.h>
#include <net/tcp_states.h>

/*
 *	Is a socket 'connection oriented' ?
 */
static inline int connection_based(struct sock *sk)
{
	return sk->sk_type == SOCK_SEQPACKET || sk->sk_type == SOCK_STREAM;
}

/*
 * Wait for a packet..
 */
static int wait_for_packet(struct sock *sk, int *err, long *timeo_p)
{
	int error;
	DEFINE_WAIT(wait);

	prepare_to_wait_exclusive(sk->sk_sleep, &wait, TASK_INTERRUPTIBLE);

	/* Socket errors? */
	error = sock_error(sk);
	if (error)
		goto out_err;

	if (!skb_queue_empty(&sk->sk_receive_queue))
		goto out;

	/* Socket shut down? */
	if (sk->sk_shutdown & RCV_SHUTDOWN)
		goto out_noerr;

	/* Sequenced packets can come disconnected.
	 * If so we report the problem
	 */
	error = -ENOTCONN;
	if (connection_based(sk) &&
	    !(sk->sk_state == TCP_ESTABLISHED || sk->sk_state == TCP_LISTEN))
		goto out_err;

	/* handle signals */
	if (signal_pending(current))
		goto interrupted;

	error = 0;
	*timeo_p = schedule_timeout(*timeo_p);
out:
	finish_wait(sk->sk_sleep, &wait);
	return error;
interrupted:
	error = sock_intr_errno(*timeo_p);
out_err:
	*err = error;
	goto out;
out_noerr:
	*err = 0;
	error = 1;
	goto out;
}

/**
 *	__skb_recv_datagram - Receive a datagram skbuff
 *	@sk: socket
 *	@flags: MSG_ flags
 *	@peeked: returns non-zero if this packet has been seen before
 *	@err: error code returned
 *
 *	Get a datagram skbuff, understands the peeking, nonblocking wakeups
 *	and possible races. This replaces identical code in packet, raw and
 *	udp, as well as the IPX AX.25 and Appletalk. It also finally fixes
 *	the long standing peek and read race for datagram sockets. If you
 *	alter this routine remember it must be re-entrant.
 *
 *	This function will lock the socket if a skb is returned, so the caller
 *	needs to unlock the socket in that case (usually by calling
 *	skb_free_datagram)
 *
 *	* It does not lock socket since today. This function is
 *	* free of race conditions. This measure should/can improve
 *	* significantly datagram socket latencies at high loads,
 *	* when data copying to user space takes lots of time.
 *	* (BTW I've just killed the last cli() in IP/IPv6/core/netlink/packet
 *	*  8) Great win.)
 *	*			                    --ANK (980729)
 *
 *	The order of the tests when we find no data waiting are specified
 *	quite explicitly by POSIX 1003.1g, don't change them without having
 *	the standard around please.
 */
struct sk_buff *__skb_recv_datagram(struct sock *sk, unsigned flags,
				    int *peeked, int *err)
{
	struct sk_buff *skb;
	long timeo;
	/*
	 * Caller is allowed not to check sk->sk_err before skb_recv_datagram()
	 */
	int error = sock_error(sk);

	if (error)
		goto no_packet;

	timeo = sock_rcvtimeo(sk, flags & MSG_DONTWAIT);

	do {
		/* Again only user level code calls this function, so nothing
		 * interrupt level will suddenly eat the receive_queue.
		 *
		 * Look at current nfs client by the way...
		 * However, this function was corrent in any case. 8)
		 */
		unsigned long cpu_flags;

		spin_lock_irqsave(&sk->sk_receive_queue.lock, cpu_flags);
		skb = skb_peek(&sk->sk_receive_queue);
		if (skb) {
			*peeked = skb->peeked;
			if (flags & MSG_PEEK) {
				skb->peeked = 1;
				atomic_inc(&skb->users);
			} else
				__skb_unlink(skb, &sk->sk_receive_queue);
		}
		spin_unlock_irqrestore(&sk->sk_receive_queue.lock, cpu_flags);

		if (skb)
			return skb;

		/* User doesn't want to wait */
		error = -EAGAIN;
		if (!timeo)
			goto no_packet;

	} while (!wait_for_packet(sk, err, &timeo));

	return NULL;

no_packet:
	*err = error;
	return NULL;
}
EXPORT_SYMBOL(__skb_recv_datagram);

struct sk_buff *skb_recv_datagram(struct sock *sk, unsigned flags,
				  int noblock, int *err)
{
	int peeked;

	return __skb_recv_datagram(sk, flags | (noblock ? MSG_DONTWAIT : 0),
				   &peeked, err);
}

void skb_free_datagram(struct sock *sk, struct sk_buff *skb)
{
	consume_skb(skb);
	sk_mem_reclaim_partial(sk);
}

/**
 *	skb_kill_datagram - Free a datagram skbuff forcibly
 *	@sk: socket
 *	@skb: datagram skbuff
 *	@flags: MSG_ flags
 *
 *	This function frees a datagram skbuff that was received by
 *	skb_recv_datagram.  The flags argument must match the one
 *	used for skb_recv_datagram.
 *
 *	If the MSG_PEEK flag is set, and the packet is still on the
 *	receive queue of the socket, it will be taken off the queue
 *	before it is freed.
 *
 *	This function currently only disables BH when acquiring the
 *	sk_receive_queue lock.  Therefore it must not be used in a
 *	context where that lock is acquired in an IRQ context.
 *
 *	It returns 0 if the packet was removed by us.
 */

int skb_kill_datagram(struct sock *sk, struct sk_buff *skb, unsigned int flags)
{
	int err = 0;

	if (flags & MSG_PEEK) {
		err = -ENOENT;
		spin_lock_bh(&sk->sk_receive_queue.lock);
		if (skb == skb_peek(&sk->sk_receive_queue)) {
			__skb_unlink(skb, &sk->sk_receive_queue);
			atomic_dec(&skb->users);
			err = 0;
		}
		spin_unlock_bh(&sk->sk_receive_queue.lock);
	}

	skb_free_datagram(sk, skb);
	return err;
}

EXPORT_SYMBOL(skb_kill_datagram);

/**
 *	skb_copy_datagram_iovec - Copy a datagram to an iovec.
 *	@skb: buffer to copy
 *	@offset: offset in the buffer to start copying from
 *	@to: io vector to copy to
 *	@len: amount of data to copy from buffer to iovec
 *
 *	Note: the iovec is modified during the copy.
 */
int skb_copy_datagram_iovec(const struct sk_buff *skb, int offset,
			    struct iovec *to, int len)
{
	int start = skb_headlen(skb);
	int i, copy = start - offset;

	/* Copy header. */
	if (copy > 0) {
		if (copy > len)
			copy = len;
		if (memcpy_toiovec(to, skb->data + offset, copy))
			goto fault;
		if ((len -= copy) == 0)
			return 0;
		offset += copy;
	}

	/* Copy paged appendix. Hmm... why does this look so complicated? */
	for (i = 0; i < skb_shinfo(skb)->nr_frags; i++) {
		int end;

		WARN_ON(start > offset + len);

		end = start + skb_shinfo(skb)->frags[i].size;
		if ((copy = end - offset) > 0) {
			int err;
			u8  *vaddr;
			skb_frag_t *frag = &skb_shinfo(skb)->frags[i];
			struct page *page = frag->page;

			if (copy > len)
				copy = len;
			vaddr = kmap(page);
			err = memcpy_toiovec(to, vaddr + frag->page_offset +
					     offset - start, copy);
			kunmap(page);
			if (err)
				goto fault;
			if (!(len -= copy))
				return 0;
			offset += copy;
		}
		start = end;
	}

	if (skb_shinfo(skb)->frag_list) {
		struct sk_buff *list = skb_shinfo(skb)->frag_list;

		for (; list; list = list->next) {
			int end;

			WARN_ON(start > offset + len);

			end = start + list->len;
			if ((copy = end - offset) > 0) {
				if (copy > len)
					copy = len;
				if (skb_copy_datagram_iovec(list,
							    offset - start,
							    to, copy))
					goto fault;
				if ((len -= copy) == 0)
					return 0;
				offset += copy;
			}
			start = end;
		}
	}
	if (!len)
		return 0;

fault:
	return -EFAULT;
}

/**
 *	skb_copy_datagram_const_iovec - Copy a datagram to an iovec.
 *	@skb: buffer to copy
 *	@offset: offset in the buffer to start copying from
 *	@to: io vector to copy to
 *	@to_offset: offset in the io vector to start copying to
 *	@len: amount of data to copy from buffer to iovec
 *
 *	Returns 0 or -EFAULT.
 *	Note: the iovec is not modified during the copy.
 */
int skb_copy_datagram_const_iovec(const struct sk_buff *skb, int offset,
				  const struct iovec *to, int to_offset,
				  int len)
{
	int start = skb_headlen(skb);
	int i, copy = start - offset;

	/* Copy header. */
	if (copy > 0) {
		if (copy > len)
			copy = len;
		if (memcpy_toiovecend(to, skb->data + offset, to_offset, copy))
			goto fault;
		if ((len -= copy) == 0)
			return 0;
		offset += copy;
		to_offset += copy;
	}

	/* Copy paged appendix. Hmm... why does this look so complicated? */
	for (i = 0; i < skb_shinfo(skb)->nr_frags; i++) {
		int end;

		WARN_ON(start > offset + len);

		end = start + skb_shinfo(skb)->frags[i].size;
		if ((copy = end - offset) > 0) {
			int err;
			u8  *vaddr;
			skb_frag_t *frag = &skb_shinfo(skb)->frags[i];
			struct page *page = frag->page;

			if (copy > len)
				copy = len;
			vaddr = kmap(page);
			err = memcpy_toiovecend(to, vaddr + frag->page_offset +
						offset - start, to_offset, copy);
			kunmap(page);
			if (err)
				goto fault;
			if (!(len -= copy))
				return 0;
			offset += copy;
			to_offset += copy;
		}
		start = end;
	}

	if (skb_shinfo(skb)->frag_list) {
		struct sk_buff *list = skb_shinfo(skb)->frag_list;

		for (; list; list = list->next) {
			int end;

			WARN_ON(start > offset + len);

			end = start + list->len;
			if ((copy = end - offset) > 0) {
				if (copy > len)
					copy = len;
				if (skb_copy_datagram_const_iovec(list,
							    offset - start,
							    to, to_offset,
							    copy))
					goto fault;
				if ((len -= copy) == 0)
					return 0;
				offset += copy;
				to_offset += copy;
			}
			start = end;
		}
	}
	if (!len)
		return 0;

fault:
	return -EFAULT;
}
EXPORT_SYMBOL(skb_copy_datagram_const_iovec);

/**
 *	skb_copy_datagram_from_iovec - Copy a datagram from an iovec.
 *	@skb: buffer to copy
 *	@offset: offset in the buffer to start copying to
 *	@from: io vector to copy to
 *	@len: amount of data to copy to buffer from iovec
 *
 *	Returns 0 or -EFAULT.
 *	Note: the iovec is modified during the copy.
 */
int skb_copy_datagram_from_iovec(struct sk_buff *skb, int offset,
				 struct iovec *from, int len)
{
	int start = skb_headlen(skb);
	int i, copy = start - offset;

	/* Copy header. */
	if (copy > 0) {
		if (copy > len)
			copy = len;
		if (memcpy_fromiovec(skb->data + offset, from, copy))
			goto fault;
		if ((len -= copy) == 0)
			return 0;
		offset += copy;
	}

	/* Copy paged appendix. Hmm... why does this look so complicated? */
	for (i = 0; i < skb_shinfo(skb)->nr_frags; i++) {
		int end;

		WARN_ON(start > offset + len);

		end = start + skb_shinfo(skb)->frags[i].size;
		if ((copy = end - offset) > 0) {
			int err;
			u8  *vaddr;
			skb_frag_t *frag = &skb_shinfo(skb)->frags[i];
			struct page *page = frag->page;

			if (copy > len)
				copy = len;
			vaddr = kmap(page);
			err = memcpy_fromiovec(vaddr + frag->page_offset +
					       offset - start, from, copy);
			kunmap(page);
			if (err)
				goto fault;

			if (!(len -= copy))
				return 0;
			offset += copy;
		}
		start = end;
	}

	if (skb_shinfo(skb)->frag_list) {
		struct sk_buff *list = skb_shinfo(skb)->frag_list;

		for (; list; list = list->next) {
			int end;

			WARN_ON(start > offset + len);

			end = start + list->len;
			if ((copy = end - offset) > 0) {
				if (copy > len)
					copy = len;
				if (skb_copy_datagram_from_iovec(list,
								 offset - start,
								 from, copy))
					goto fault;
				if ((len -= copy) == 0)
					return 0;
				offset += copy;
			}
			start = end;
		}
	}
	if (!len)
		return 0;

fault:
	return -EFAULT;
}
EXPORT_SYMBOL(skb_copy_datagram_from_iovec);

static int skb_copy_and_csum_datagram(const struct sk_buff *skb, int offset,
				      u8 __user *to, int len,
				      __wsum *csump)
{
	int start = skb_headlen(skb);
	int pos = 0;
	int i, copy = start - offset;

	/* Copy header. */
	if (copy > 0) {
		int err = 0;
		if (copy > len)
			copy = len;
		*csump = csum_and_copy_to_user(skb->data + offset, to, copy,
					       *csump, &err);
		if (err)
			goto fault;
		if ((len -= copy) == 0)
			return 0;
		offset += copy;
		to += copy;
		pos = copy;
	}

	for (i = 0; i < skb_shinfo(skb)->nr_frags; i++) {
		int end;

		WARN_ON(start > offset + len);

		end = start + skb_shinfo(skb)->frags[i].size;
		if ((copy = end - offset) > 0) {
			__wsum csum2;
			int err = 0;
			u8  *vaddr;
			skb_frag_t *frag = &skb_shinfo(skb)->frags[i];
			struct page *page = frag->page;

			if (copy > len)
				copy = len;
			vaddr = kmap(page);
			csum2 = csum_and_copy_to_user(vaddr +
							frag->page_offset +
							offset - start,
						      to, copy, 0, &err);
			kunmap(page);
			if (err)
				goto fault;
			*csump = csum_block_add(*csump, csum2, pos);
			if (!(len -= copy))
				return 0;
			offset += copy;
			to += copy;
			pos += copy;
		}
		start = end;
	}

	if (skb_shinfo(skb)->frag_list) {
		struct sk_buff *list = skb_shinfo(skb)->frag_list;

		for (; list; list=list->next) {
			int end;

			WARN_ON(start > offset + len);

			end = start + list->len;
			if ((copy = end - offset) > 0) {
				__wsum csum2 = 0;
				if (copy > len)
					copy = len;
				if (skb_copy_and_csum_datagram(list,
							       offset - start,
							       to, copy,
							       &csum2))
					goto fault;
				*csump = csum_block_add(*csump, csum2, pos);
				if ((len -= copy) == 0)
					return 0;
				offset += copy;
				to += copy;
				pos += copy;
			}
			start = end;
		}
	}
	if (!len)
		return 0;

fault:
	return -EFAULT;
}

__sum16 __skb_checksum_complete_head(struct sk_buff *skb, int len)
{
	__sum16 sum;

	sum = csum_fold(skb_checksum(skb, 0, len, skb->csum));
	if (likely(!sum)) {
		if (unlikely(skb->ip_summed == CHECKSUM_COMPLETE))
			netdev_rx_csum_fault(skb->dev);
		skb->ip_summed = CHECKSUM_UNNECESSARY;
	}
	return sum;
}
EXPORT_SYMBOL(__skb_checksum_complete_head);

__sum16 __skb_checksum_complete(struct sk_buff *skb)
{
	return __skb_checksum_complete_head(skb, skb->len);
}
EXPORT_SYMBOL(__skb_checksum_complete);

/**
 *	skb_copy_and_csum_datagram_iovec - Copy and checkum skb to user iovec.
 *	@skb: skbuff
 *	@hlen: hardware length
 *	@iov: io vector
 *
 *	Caller _must_ check that skb will fit to this iovec.
 *
 *	Returns: 0       - success.
 *		 -EINVAL - checksum failure.
 *		 -EFAULT - fault during copy. Beware, in this case iovec
 *			   can be modified!
 */
int skb_copy_and_csum_datagram_iovec(struct sk_buff *skb,
				     int hlen, struct iovec *iov)
{
	__wsum csum;
	int chunk = skb->len - hlen;

	if (!chunk)
		return 0;

	/* Skip filled elements.
	 * Pretty silly, look at memcpy_toiovec, though 8)
	 */
	while (!iov->iov_len)
		iov++;

	if (iov->iov_len < chunk) {
		if (__skb_checksum_complete(skb))
			goto csum_error;
		if (skb_copy_datagram_iovec(skb, hlen, iov, chunk))
			goto fault;
	} else {
		csum = csum_partial(skb->data, hlen, skb->csum);
		if (skb_copy_and_csum_datagram(skb, hlen, iov->iov_base,
					       chunk, &csum))
			goto fault;
		if (csum_fold(csum))
			goto csum_error;
		if (unlikely(skb->ip_summed == CHECKSUM_COMPLETE))
			netdev_rx_csum_fault(skb->dev);
		iov->iov_len -= chunk;
		iov->iov_base += chunk;
	}
	return 0;
csum_error:
	return -EINVAL;
fault:
	return -EFAULT;
}

/**
 * 	datagram_poll - generic datagram poll
 *	@file: file struct
 *	@sock: socket
 *	@wait: poll table
 *
 *	Datagram poll: Again totally generic. This also handles
 *	sequenced packet sockets providing the socket receive queue
 *	is only ever holding data ready to receive.
 *
 *	Note: when you _don't_ use this routine for this protocol,
 *	and you use a different write policy from sock_writeable()
 *	then please supply your own write_space callback.
 */
unsigned int datagram_poll(struct file *file, struct socket *sock,
			   poll_table *wait)
{
	struct sock *sk = sock->sk;
	unsigned int mask;

	poll_wait(file, sk->sk_sleep, wait);
	mask = 0;

	/* exceptional events? */
	if (sk->sk_err || !skb_queue_empty(&sk->sk_error_queue))
		mask |= POLLERR;
	if (sk->sk_shutdown & RCV_SHUTDOWN)
		mask |= POLLRDHUP;
	if (sk->sk_shutdown == SHUTDOWN_MASK)
		mask |= POLLHUP;

	/* readable? */
	if (!skb_queue_empty(&sk->sk_receive_queue) ||
	    (sk->sk_shutdown & RCV_SHUTDOWN))
		mask |= POLLIN | POLLRDNORM;

	/* Connection-based need to check for termination and startup */
	if (connection_based(sk)) {
		if (sk->sk_state == TCP_CLOSE)
			mask |= POLLHUP;
		/* connection hasn't started yet? */
		if (sk->sk_state == TCP_SYN_SENT)
			return mask;
	}

	/* writable? */
	if (sock_writeable(sk))
		mask |= POLLOUT | POLLWRNORM | POLLWRBAND;
	else
		set_bit(SOCK_ASYNC_NOSPACE, &sk->sk_socket->flags);

	return mask;
}

EXPORT_SYMBOL(datagram_poll);
EXPORT_SYMBOL(skb_copy_and_csum_datagram_iovec);
EXPORT_SYMBOL(skb_copy_datagram_iovec);
EXPORT_SYMBOL(skb_free_datagram);
EXPORT_SYMBOL(skb_recv_datagram);
Commit	Line	Data
1da177e4 LT	1	/*
	2	* SUCS NET3:
	3	*
	4	* Generic datagram handling routines. These are generic for all
	5	* protocols. Possibly a generic IP version on top of these would
	6	* make sense. Not tonight however 8-).
	7	* This is used because UDP, RAW, PACKET, DDP, IPX, AX.25 and
	8	* NetROM layer all have identical poll code and mostly
	9	* identical recvmsg() code. So we share it here. The poll was
	10	* shared before but buried in udp.c so I moved it.
	11	*
113aa838	12	* Authors: Alan Cox <alan@lxorguk.ukuu.org.uk>. (datagram_poll() from old
1da177e4 LT	13	* udp.c code)
	14	*
	15	* Fixes:
	16	* Alan Cox : NULL return from skb_peek_copy()
	17	* understood
	18	* Alan Cox : Rewrote skb_read_datagram to avoid the
	19	* skb_peek_copy stuff.
	20	* Alan Cox : Added support for SOCK_SEQPACKET.
	21	* IPX can no longer use the SO_TYPE hack
	22	* but AX.25 now works right, and SPX is
	23	* feasible.
	24	* Alan Cox : Fixed write poll of non IP protocol
	25	* crash.
	26	* Florian La Roche: Changed for my new skbuff handling.
	27	* Darryl Miles : Fixed non-blocking SOCK_SEQPACKET.
	28	* Linus Torvalds : BSD semantic fixes.
	29	* Alan Cox : Datagram iovec handling
	30	* Darryl Miles : Fixed non-blocking SOCK_STREAM.
	31	* Alan Cox : POSIXisms
	32	* Pete Wyckoff : Unconnected accept() fix.
	33	*
	34	*/
	35
	36	#include <linux/module.h>
	37	#include <linux/types.h>
	38	#include <linux/kernel.h>
	39	#include <asm/uaccess.h>
	40	#include <asm/system.h>
	41	#include <linux/mm.h>
	42	#include <linux/interrupt.h>
	43	#include <linux/errno.h>
	44	#include <linux/sched.h>
	45	#include <linux/inet.h>
1da177e4 LT	46	#include <linux/netdevice.h>
	47	#include <linux/rtnetlink.h>
	48	#include <linux/poll.h>
	49	#include <linux/highmem.h>
3305b80c	50	#include <linux/spinlock.h>
1da177e4 LT	51
	52	#include <net/protocol.h>
	53	#include <linux/skbuff.h>
1da177e4	54
c752f073 ACM	55	#include <net/checksum.h>
	56	#include <net/sock.h>
	57	#include <net/tcp_states.h>
1da177e4 LT	58
	59	/*
	60	* Is a socket 'connection oriented' ?
	61	*/
	62	static inline int connection_based(struct sock *sk)
	63	{
	64	return sk->sk_type == SOCK_SEQPACKET \|\| sk->sk_type == SOCK_STREAM;
	65	}
	66
	67	/*
	68	* Wait for a packet..
	69	*/
	70	static int wait_for_packet(struct sock sk, int err, long *timeo_p)
	71	{
	72	int error;
	73	DEFINE_WAIT(wait);
	74
	75	prepare_to_wait_exclusive(sk->sk_sleep, &wait, TASK_INTERRUPTIBLE);
	76
	77	/* Socket errors? */
	78	error = sock_error(sk);
	79	if (error)
	80	goto out_err;
	81
	82	if (!skb_queue_empty(&sk->sk_receive_queue))
	83	goto out;
	84
	85	/* Socket shut down? */
	86	if (sk->sk_shutdown & RCV_SHUTDOWN)
	87	goto out_noerr;
	88
	89	/* Sequenced packets can come disconnected.
	90	* If so we report the problem
	91	*/
	92	error = -ENOTCONN;
	93	if (connection_based(sk) &&
	94	!(sk->sk_state == TCP_ESTABLISHED \|\| sk->sk_state == TCP_LISTEN))
	95	goto out_err;
	96
	97	/* handle signals */
	98	if (signal_pending(current))
	99	goto interrupted;
	100
	101	error = 0;
	102	timeo_p = schedule_timeout(timeo_p);
	103	out:
	104	finish_wait(sk->sk_sleep, &wait);
	105	return error;
	106	interrupted:
	107	error = sock_intr_errno(*timeo_p);
	108	out_err:
	109	*err = error;
	110	goto out;
	111	out_noerr:
	112	*err = 0;
	113	error = 1;
	114	goto out;
	115	}
	116
	117	/**
a59322be	118	* __skb_recv_datagram - Receive a datagram skbuff
4dc3b16b PP	119	* @sk: socket
4dc3b16b PP	120	* @flags: MSG_ flags
a59322be	121	* @peeked: returns non-zero if this packet has been seen before
4dc3b16b	122	* @err: error code returned
1da177e4 LT	123	*
	124	* Get a datagram skbuff, understands the peeking, nonblocking wakeups
	125	* and possible races. This replaces identical code in packet, raw and
	126	* udp, as well as the IPX AX.25 and Appletalk. It also finally fixes
	127	* the long standing peek and read race for datagram sockets. If you
	128	* alter this routine remember it must be re-entrant.
	129	*
	130	* This function will lock the socket if a skb is returned, so the caller
	131	* needs to unlock the socket in that case (usually by calling
	132	* skb_free_datagram)
	133	*
	134	* * It does not lock socket since today. This function is
	135	* * free of race conditions. This measure should/can improve
	136	* * significantly datagram socket latencies at high loads,
	137	* * when data copying to user space takes lots of time.
	138	* * (BTW I've just killed the last cli() in IP/IPv6/core/netlink/packet
	139	* * 8) Great win.)
	140	* * --ANK (980729)
	141	*
	142	* The order of the tests when we find no data waiting are specified
	143	* quite explicitly by POSIX 1003.1g, don't change them without having
	144	* the standard around please.
	145	*/
a59322be HX	146	struct sk_buff __skb_recv_datagram(struct sock sk, unsigned flags,
a59322be HX	147	int peeked, int err)
1da177e4 LT	148	{
	149	struct sk_buff *skb;
	150	long timeo;
	151	/*
	152	* Caller is allowed not to check sk->sk_err before skb_recv_datagram()
	153	*/
	154	int error = sock_error(sk);
	155
	156	if (error)
	157	goto no_packet;
	158
a59322be	159	timeo = sock_rcvtimeo(sk, flags & MSG_DONTWAIT);
1da177e4 LT	160
	161	do {
	162	/* Again only user level code calls this function, so nothing
	163	* interrupt level will suddenly eat the receive_queue.
	164	*
	165	* Look at current nfs client by the way...
	166	* However, this function was corrent in any case. 8)
	167	*/
a59322be HX	168	unsigned long cpu_flags;
	169
	170	spin_lock_irqsave(&sk->sk_receive_queue.lock, cpu_flags);
	171	skb = skb_peek(&sk->sk_receive_queue);
	172	if (skb) {
	173	*peeked = skb->peeked;
	174	if (flags & MSG_PEEK) {
	175	skb->peeked = 1;
1da177e4	176	atomic_inc(&skb->users);
a59322be HX	177	} else
	178	__skb_unlink(skb, &sk->sk_receive_queue);
	179	}
	180	spin_unlock_irqrestore(&sk->sk_receive_queue.lock, cpu_flags);
1da177e4 LT	181
	182	if (skb)
	183	return skb;
	184
	185	/* User doesn't want to wait */
	186	error = -EAGAIN;
	187	if (!timeo)
	188	goto no_packet;
	189
	190	} while (!wait_for_packet(sk, err, &timeo));
	191
	192	return NULL;
	193
	194	no_packet:
	195	*err = error;
	196	return NULL;
	197	}
a59322be HX	198	EXPORT_SYMBOL(__skb_recv_datagram);
	199
	200	struct sk_buff skb_recv_datagram(struct sock sk, unsigned flags,
	201	int noblock, int *err)
	202	{
	203	int peeked;
	204
	205	return __skb_recv_datagram(sk, flags \| (noblock ? MSG_DONTWAIT : 0),
	206	&peeked, err);
	207	}
1da177e4 LT	208
	209	void skb_free_datagram(struct sock sk, struct sk_buff skb)
	210	{
ead2ceb0	211	consume_skb(skb);
270acefa	212	sk_mem_reclaim_partial(sk);
1da177e4 LT	213	}
1da177e4 LT	214
3305b80c HX	215	/**
	216	* skb_kill_datagram - Free a datagram skbuff forcibly
	217	* @sk: socket
	218	* @skb: datagram skbuff
	219	* @flags: MSG_ flags
	220	*
	221	* This function frees a datagram skbuff that was received by
	222	* skb_recv_datagram. The flags argument must match the one
	223	* used for skb_recv_datagram.
	224	*
	225	* If the MSG_PEEK flag is set, and the packet is still on the
	226	* receive queue of the socket, it will be taken off the queue
	227	* before it is freed.
	228	*
	229	* This function currently only disables BH when acquiring the
	230	* sk_receive_queue lock. Therefore it must not be used in a
	231	* context where that lock is acquired in an IRQ context.
27ab2568 HX	232	*
27ab2568 HX	233	* It returns 0 if the packet was removed by us.
3305b80c HX	234	*/
3305b80c HX	235
27ab2568	236	int skb_kill_datagram(struct sock sk, struct sk_buff skb, unsigned int flags)
3305b80c	237	{
27ab2568 HX	238	int err = 0;
27ab2568 HX	239
3305b80c	240	if (flags & MSG_PEEK) {
27ab2568	241	err = -ENOENT;
3305b80c HX	242	spin_lock_bh(&sk->sk_receive_queue.lock);
	243	if (skb == skb_peek(&sk->sk_receive_queue)) {
	244	__skb_unlink(skb, &sk->sk_receive_queue);
	245	atomic_dec(&skb->users);
27ab2568	246	err = 0;
3305b80c HX	247	}
	248	spin_unlock_bh(&sk->sk_receive_queue.lock);
	249	}
	250
270acefa	251	skb_free_datagram(sk, skb);
27ab2568	252	return err;
3305b80c HX	253	}
	254
	255	EXPORT_SYMBOL(skb_kill_datagram);
	256
1da177e4 LT	257	/**
1da177e4 LT	258	* skb_copy_datagram_iovec - Copy a datagram to an iovec.
4dc3b16b PP	259	* @skb: buffer to copy
4dc3b16b PP	260	* @offset: offset in the buffer to start copying from
67be2dd1	261	* @to: io vector to copy to
4dc3b16b	262	* @len: amount of data to copy from buffer to iovec
1da177e4 LT	263	*
	264	* Note: the iovec is modified during the copy.
	265	*/
	266	int skb_copy_datagram_iovec(const struct sk_buff *skb, int offset,
	267	struct iovec *to, int len)
	268	{
1a028e50 DM	269	int start = skb_headlen(skb);
1a028e50 DM	270	int i, copy = start - offset;
c75d721c	271
b4d9eda0 DM	272	/* Copy header. */
	273	if (copy > 0) {
	274	if (copy > len)
	275	copy = len;
	276	if (memcpy_toiovec(to, skb->data + offset, copy))
	277	goto fault;
	278	if ((len -= copy) == 0)
	279	return 0;
	280	offset += copy;
	281	}
c75d721c	282
b4d9eda0 DM	283	/* Copy paged appendix. Hmm... why does this look so complicated? */
b4d9eda0 DM	284	for (i = 0; i < skb_shinfo(skb)->nr_frags; i++) {
1a028e50	285	int end;
1da177e4	286
547b792c	287	WARN_ON(start > offset + len);
1a028e50 DM	288
1a028e50 DM	289	end = start + skb_shinfo(skb)->frags[i].size;
b4d9eda0 DM	290	if ((copy = end - offset) > 0) {
	291	int err;
	292	u8 *vaddr;
	293	skb_frag_t *frag = &skb_shinfo(skb)->frags[i];
	294	struct page *page = frag->page;
1da177e4 LT	295
	296	if (copy > len)
	297	copy = len;
b4d9eda0	298	vaddr = kmap(page);
1a028e50 DM	299	err = memcpy_toiovec(to, vaddr + frag->page_offset +
1a028e50 DM	300	offset - start, copy);
b4d9eda0	301	kunmap(page);
1da177e4 LT	302	if (err)
	303	goto fault;
	304	if (!(len -= copy))
	305	return 0;
	306	offset += copy;
	307	}
1a028e50	308	start = end;
1da177e4	309	}
b4d9eda0 DM	310
	311	if (skb_shinfo(skb)->frag_list) {
	312	struct sk_buff *list = skb_shinfo(skb)->frag_list;
	313
	314	for (; list; list = list->next) {
1a028e50 DM	315	int end;
1a028e50 DM	316
547b792c	317	WARN_ON(start > offset + len);
b4d9eda0	318
1a028e50	319	end = start + list->len;
b4d9eda0 DM	320	if ((copy = end - offset) > 0) {
	321	if (copy > len)
	322	copy = len;
1a028e50 DM	323	if (skb_copy_datagram_iovec(list,
	324	offset - start,
	325	to, copy))
b4d9eda0 DM	326	goto fault;
	327	if ((len -= copy) == 0)
	328	return 0;
	329	offset += copy;
	330	}
1a028e50	331	start = end;
b4d9eda0	332	}
1da177e4	333	}
b4d9eda0 DM	334	if (!len)
	335	return 0;
	336
1da177e4 LT	337	fault:
	338	return -EFAULT;
	339	}
	340
0a1ec07a MT	341	/**
	342	* skb_copy_datagram_const_iovec - Copy a datagram to an iovec.
	343	* @skb: buffer to copy
	344	* @offset: offset in the buffer to start copying from
	345	* @to: io vector to copy to
	346	* @to_offset: offset in the io vector to start copying to
	347	* @len: amount of data to copy from buffer to iovec
	348	*
	349	* Returns 0 or -EFAULT.
	350	* Note: the iovec is not modified during the copy.
	351	*/
	352	int skb_copy_datagram_const_iovec(const struct sk_buff *skb, int offset,
	353	const struct iovec *to, int to_offset,
	354	int len)
	355	{
	356	int start = skb_headlen(skb);
	357	int i, copy = start - offset;
	358
	359	/* Copy header. */
	360	if (copy > 0) {
	361	if (copy > len)
	362	copy = len;
	363	if (memcpy_toiovecend(to, skb->data + offset, to_offset, copy))
	364	goto fault;
	365	if ((len -= copy) == 0)
	366	return 0;
	367	offset += copy;
	368	to_offset += copy;
	369	}
	370
	371	/* Copy paged appendix. Hmm... why does this look so complicated? */
	372	for (i = 0; i < skb_shinfo(skb)->nr_frags; i++) {
	373	int end;
	374
	375	WARN_ON(start > offset + len);
	376
	377	end = start + skb_shinfo(skb)->frags[i].size;
	378	if ((copy = end - offset) > 0) {
	379	int err;
	380	u8 *vaddr;
	381	skb_frag_t *frag = &skb_shinfo(skb)->frags[i];
	382	struct page *page = frag->page;
	383
	384	if (copy > len)
	385	copy = len;
	386	vaddr = kmap(page);
	387	err = memcpy_toiovecend(to, vaddr + frag->page_offset +
	388	offset - start, to_offset, copy);
	389	kunmap(page);
	390	if (err)
	391	goto fault;
	392	if (!(len -= copy))
	393	return 0;
	394	offset += copy;
	395	to_offset += copy;
	396	}
	397	start = end;
	398	}
	399
	400	if (skb_shinfo(skb)->frag_list) {
	401	struct sk_buff *list = skb_shinfo(skb)->frag_list;
	402
	403	for (; list; list = list->next) {
	404	int end;
405
406	WARN_ON(start > offset + len);
407
408	end = start + list->len;
409	if ((copy = end - offset) > 0) {
410	if (copy > len)
411	copy = len;
412	if (skb_copy_datagram_const_iovec(list,
413	offset - start,
414	to, to_offset,
415	copy))
416	goto fault;
417	if ((len -= copy) == 0)
418	return 0;
419	offset += copy;
420	to_offset += copy;
421	}
422	start = end;
423	}
424	}
425	if (!len)
426	return 0;
427
428	fault:
429	return -EFAULT;
430	}
431	EXPORT_SYMBOL(skb_copy_datagram_const_iovec);
432
db543c1f RR	433	/**
	434	* skb_copy_datagram_from_iovec - Copy a datagram from an iovec.
	435	* @skb: buffer to copy
	436	* @offset: offset in the buffer to start copying to
	437	* @from: io vector to copy to
	438	* @len: amount of data to copy to buffer from iovec
	439	*
	440	* Returns 0 or -EFAULT.
	441	* Note: the iovec is modified during the copy.
	442	*/
	443	int skb_copy_datagram_from_iovec(struct sk_buff *skb, int offset,
	444	struct iovec *from, int len)
	445	{
	446	int start = skb_headlen(skb);
	447	int i, copy = start - offset;
	448
	449	/* Copy header. */
	450	if (copy > 0) {
	451	if (copy > len)
	452	copy = len;
	453	if (memcpy_fromiovec(skb->data + offset, from, copy))
	454	goto fault;
	455	if ((len -= copy) == 0)
	456	return 0;
	457	offset += copy;
	458	}
	459
	460	/* Copy paged appendix. Hmm... why does this look so complicated? */
	461	for (i = 0; i < skb_shinfo(skb)->nr_frags; i++) {
	462	int end;
	463
	464	WARN_ON(start > offset + len);
	465
	466	end = start + skb_shinfo(skb)->frags[i].size;
	467	if ((copy = end - offset) > 0) {
	468	int err;
	469	u8 *vaddr;
	470	skb_frag_t *frag = &skb_shinfo(skb)->frags[i];
	471	struct page *page = frag->page;
	472
	473	if (copy > len)
	474	copy = len;
	475	vaddr = kmap(page);
	476	err = memcpy_fromiovec(vaddr + frag->page_offset +
	477	offset - start, from, copy);
	478	kunmap(page);
	479	if (err)
	480	goto fault;
	481
	482	if (!(len -= copy))
	483	return 0;
	484	offset += copy;
	485	}
	486	start = end;
	487	}
	488
	489	if (skb_shinfo(skb)->frag_list) {
	490	struct sk_buff *list = skb_shinfo(skb)->frag_list;
	491
	492	for (; list; list = list->next) {
	493	int end;
	494
	495	WARN_ON(start > offset + len);
	496
497	end = start + list->len;
498	if ((copy = end - offset) > 0) {
499	if (copy > len)
500	copy = len;
501	if (skb_copy_datagram_from_iovec(list,
502	offset - start,
503	from, copy))
504	goto fault;
505	if ((len -= copy) == 0)
506	return 0;
507	offset += copy;
508	}
509	start = end;
510	}
511	}
512	if (!len)
513	return 0;
514
515	fault:
516	return -EFAULT;
517	}
518	EXPORT_SYMBOL(skb_copy_datagram_from_iovec);
519
1da177e4 LT	520	static int skb_copy_and_csum_datagram(const struct sk_buff *skb, int offset,
1da177e4 LT	521	u8 __user *to, int len,
5084205f	522	__wsum *csump)
1da177e4	523	{
1a028e50	524	int start = skb_headlen(skb);
1da177e4	525	int pos = 0;
1a028e50	526	int i, copy = start - offset;
1da177e4 LT	527
	528	/* Copy header. */
	529	if (copy > 0) {
	530	int err = 0;
	531	if (copy > len)
	532	copy = len;
	533	*csump = csum_and_copy_to_user(skb->data + offset, to, copy,
	534	*csump, &err);
	535	if (err)
	536	goto fault;
	537	if ((len -= copy) == 0)
	538	return 0;
	539	offset += copy;
	540	to += copy;
	541	pos = copy;
	542	}
	543
	544	for (i = 0; i < skb_shinfo(skb)->nr_frags; i++) {
1a028e50	545	int end;
1da177e4	546
547b792c	547	WARN_ON(start > offset + len);
1a028e50 DM	548
1a028e50 DM	549	end = start + skb_shinfo(skb)->frags[i].size;
1da177e4	550	if ((copy = end - offset) > 0) {
5084205f	551	__wsum csum2;
1da177e4 LT	552	int err = 0;
	553	u8 *vaddr;
	554	skb_frag_t *frag = &skb_shinfo(skb)->frags[i];
	555	struct page *page = frag->page;
	556
	557	if (copy > len)
	558	copy = len;
	559	vaddr = kmap(page);
	560	csum2 = csum_and_copy_to_user(vaddr +
1a028e50 DM	561	frag->page_offset +
1a028e50 DM	562	offset - start,
1da177e4 LT	563	to, copy, 0, &err);
	564	kunmap(page);
	565	if (err)
	566	goto fault;
	567	csump = csum_block_add(csump, csum2, pos);
	568	if (!(len -= copy))
	569	return 0;
	570	offset += copy;
	571	to += copy;
	572	pos += copy;
	573	}
1a028e50	574	start = end;
1da177e4 LT	575	}
	576
	577	if (skb_shinfo(skb)->frag_list) {
	578	struct sk_buff *list = skb_shinfo(skb)->frag_list;
	579
	580	for (; list; list=list->next) {
1a028e50 DM	581	int end;
1a028e50 DM	582
547b792c	583	WARN_ON(start > offset + len);
1da177e4	584
1a028e50	585	end = start + list->len;
1da177e4	586	if ((copy = end - offset) > 0) {
5084205f	587	__wsum csum2 = 0;
1da177e4 LT	588	if (copy > len)
1da177e4 LT	589	copy = len;
1a028e50 DM	590	if (skb_copy_and_csum_datagram(list,
1a028e50 DM	591	offset - start,
1da177e4 LT	592	to, copy,
	593	&csum2))
	594	goto fault;
	595	csump = csum_block_add(csump, csum2, pos);
	596	if ((len -= copy) == 0)
	597	return 0;
	598	offset += copy;
	599	to += copy;
	600	pos += copy;
	601	}
1a028e50	602	start = end;
1da177e4 LT	603	}
	604	}
	605	if (!len)
	606	return 0;
	607
	608	fault:
	609	return -EFAULT;
	610	}
	611
759e5d00	612	__sum16 __skb_checksum_complete_head(struct sk_buff *skb, int len)
fb286bb2	613	{
d3bc23e7	614	__sum16 sum;
fb286bb2	615
759e5d00	616	sum = csum_fold(skb_checksum(skb, 0, len, skb->csum));
fb286bb2	617	if (likely(!sum)) {
84fa7933	618	if (unlikely(skb->ip_summed == CHECKSUM_COMPLETE))
fb286bb2 HX	619	netdev_rx_csum_fault(skb->dev);
	620	skb->ip_summed = CHECKSUM_UNNECESSARY;
	621	}
	622	return sum;
	623	}
759e5d00 HX	624	EXPORT_SYMBOL(__skb_checksum_complete_head);
	625
	626	__sum16 __skb_checksum_complete(struct sk_buff *skb)
	627	{
	628	return __skb_checksum_complete_head(skb, skb->len);
	629	}
fb286bb2 HX	630	EXPORT_SYMBOL(__skb_checksum_complete);
fb286bb2 HX	631
1da177e4 LT	632	/**
1da177e4 LT	633	* skb_copy_and_csum_datagram_iovec - Copy and checkum skb to user iovec.
4dc3b16b PP	634	* @skb: skbuff
4dc3b16b PP	635	* @hlen: hardware length
67be2dd1	636	* @iov: io vector
4ec93edb	637	*
1da177e4 LT	638	* Caller _must_ check that skb will fit to this iovec.
	639	*
	640	* Returns: 0 - success.
	641	* -EINVAL - checksum failure.
	642	* -EFAULT - fault during copy. Beware, in this case iovec
	643	* can be modified!
	644	*/
fb286bb2	645	int skb_copy_and_csum_datagram_iovec(struct sk_buff *skb,
1da177e4 LT	646	int hlen, struct iovec *iov)
1da177e4 LT	647	{
d3bc23e7	648	__wsum csum;
1da177e4 LT	649	int chunk = skb->len - hlen;
1da177e4 LT	650
ef8aef55 HX	651	if (!chunk)
	652	return 0;
	653
1da177e4 LT	654	/* Skip filled elements.
	655	* Pretty silly, look at memcpy_toiovec, though 8)
	656	*/
	657	while (!iov->iov_len)
	658	iov++;
	659
	660	if (iov->iov_len < chunk) {
fb286bb2	661	if (__skb_checksum_complete(skb))
1da177e4 LT	662	goto csum_error;
	663	if (skb_copy_datagram_iovec(skb, hlen, iov, chunk))
	664	goto fault;
	665	} else {
	666	csum = csum_partial(skb->data, hlen, skb->csum);
	667	if (skb_copy_and_csum_datagram(skb, hlen, iov->iov_base,
	668	chunk, &csum))
	669	goto fault;
d3bc23e7	670	if (csum_fold(csum))
1da177e4	671	goto csum_error;
84fa7933	672	if (unlikely(skb->ip_summed == CHECKSUM_COMPLETE))
fb286bb2	673	netdev_rx_csum_fault(skb->dev);
1da177e4 LT	674	iov->iov_len -= chunk;
	675	iov->iov_base += chunk;
	676	}
	677	return 0;
	678	csum_error:
	679	return -EINVAL;
	680	fault:
	681	return -EFAULT;
	682	}
	683
	684	/**
	685	* datagram_poll - generic datagram poll
4dc3b16b PP	686	* @file: file struct
	687	* @sock: socket
	688	* @wait: poll table
1da177e4 LT	689	*
	690	* Datagram poll: Again totally generic. This also handles
	691	* sequenced packet sockets providing the socket receive queue
	692	* is only ever holding data ready to receive.
	693	*
	694	* Note: when you _don't_ use this routine for this protocol,
	695	* and you use a different write policy from sock_writeable()
	696	* then please supply your own write_space callback.
	697	*/
	698	unsigned int datagram_poll(struct file file, struct socket sock,
	699	poll_table *wait)
	700	{
	701	struct sock *sk = sock->sk;
	702	unsigned int mask;
	703
	704	poll_wait(file, sk->sk_sleep, wait);
	705	mask = 0;
	706
	707	/* exceptional events? */
	708	if (sk->sk_err \|\| !skb_queue_empty(&sk->sk_error_queue))
	709	mask \|= POLLERR;
f348d70a DL	710	if (sk->sk_shutdown & RCV_SHUTDOWN)
f348d70a DL	711	mask \|= POLLRDHUP;
1da177e4 LT	712	if (sk->sk_shutdown == SHUTDOWN_MASK)
	713	mask \|= POLLHUP;
	714
	715	/* readable? */
	716	if (!skb_queue_empty(&sk->sk_receive_queue) \|\|
	717	(sk->sk_shutdown & RCV_SHUTDOWN))
	718	mask \|= POLLIN \| POLLRDNORM;
	719
	720	/* Connection-based need to check for termination and startup */
	721	if (connection_based(sk)) {
	722	if (sk->sk_state == TCP_CLOSE)
	723	mask \|= POLLHUP;
	724	/* connection hasn't started yet? */
	725	if (sk->sk_state == TCP_SYN_SENT)
	726	return mask;
	727	}
	728
	729	/* writable? */
	730	if (sock_writeable(sk))
	731	mask \|= POLLOUT \| POLLWRNORM \| POLLWRBAND;
	732	else
	733	set_bit(SOCK_ASYNC_NOSPACE, &sk->sk_socket->flags);
	734
	735	return mask;
	736	}
	737
	738	EXPORT_SYMBOL(datagram_poll);
	739	EXPORT_SYMBOL(skb_copy_and_csum_datagram_iovec);
	740	EXPORT_SYMBOL(skb_copy_datagram_iovec);
	741	EXPORT_SYMBOL(skb_free_datagram);
	742	EXPORT_SYMBOL(skb_recv_datagram);